Dynamic Trust Management: Network Profiling for High Assurance Resilience

Abstract

Trust Management (TM) systems are infrastructures that support efficient and secure access to resources in large decentralized systems. They provide a language for expressing authorizations and access control policies as well as a trust management engine that processes requests, to automatically address access requests. Traditionally, the enforcement of Trust Management decisions is static and involves the use of appropriate cryptographic mechanisms. However, recently two TM systems were proposed for which the enforcement is dynamic. Dynamic TM systems expand, (i) the expressibility of a system language to capture anomaly-triggered access control policies, and (ii) the enforcement capabilities via graduated response mechanisms such as Rollback Access control. These mechanisms are proactively triggered under the perceived potential of an attack: they selectively disrupt the TM-granted access to a resource temporarily, to mitigate the system threat. In this Chapter we discuss the use of real-time stochastic analyzers and graduated response security mechanisms to detect/prevent anomalies in TM systems, and propose an architecture for dynamic Trust Management that tolerates 0-day attacks and insider attacks.